Modeling, identification, and compensation control of pneumatic muscle hysteresis based on an improved generalized Bouc-Wen model

Due to the merits of high flexibility, great pulling force, and light weight, the pneumatic muscle (PM) is applied to realize the flexible driving control commonly. However, the inherent asymmetric hysteresis nonlinear characteristics affect its control accuracy seriously and tradition generalized Bouc-Wen (GBW) model and its variants cannot characterize the hysteresis of PM effectively. A novel improved generalized Bouc-Wen (IGBW) model is proposed based on the concave convex features of the PM hysteresis curve. The parameter identification effects of Levenberg-Marquardt algorithm (LM), gradient descent method, Newton iteration method, differential evolution algorithm (DE), particle swarm optimization algorithm (PSO), and genetic algorithm (GA) on the IGBW model are studied firstly. Then, the modeling capacity of IGBW model is checked by contrasting the modeling accuracy and identification efficiency of IGBW and modified symmetrical generalized Prandtl-Ishlinskii (MSGPI) models. Finally, the hysteresis compensation control strategy is designed based on the inverse model of IGBW, and its stability is analyzed. The experimental results manifest that the LM algorithm is much better than the other five estimation methods with regard to computational efficiency and identification accuracy. The IGBW model can illustrate the asymmetric hysteresis properties of PM with higher accuracy and efficiency, and the control behavior is ameliorated significantly after using the IGBW model.